|
Abstract:
The study of grammatical encoding during sentence production
has been illuminated by recent research on the occurrence and
distribution of subject-verb agreement errors. Previous experiments
on subject- verb agreement have shown that the agreement process
may be disrupted by the presence of a plural nonhead noun; e.g.,
"The article by the senators were informative" (e.g. Bock &
Miller, 1991). In general, interference may arise if a plural
nonhead appears within a modifying phrase or clause following the
head noun.
Possessive NPs present an interesting case because they are
number- bearing nonhead nouns which may appear before the head
noun. In this poster, we present three studies which examine
whether a plural possessive NP causes interference in the agreement
process.
EXPERIMENTS 1 & 2. Sentence preambles consisted of a
singular head modified by both a possessor and a prepositional
phrase (containing a "local" noun). The number specification of the
two nonheads was manipulated. In Exp. 1, preambles were presented
auditorily; and only nouns with distinct singular vs. plural
possessive forms were used (e.g., elf's vs. elves'). In Exp. 2, we
used visual presentation, and so could increase the number of items
to include orthographically distinct cases such as "city's" vs.
"cities'".
In Experiment 1, only the effect of local noun number was
significant, though there is a numerical difference in possessor NP
number. In Experiment 2, the error rates were higher (due to visual
presentation). Results of Experiment 2 show a small but significant
effect of possessor number, and a robust effect of local noun
number. (It is noteworthy how little effect the possessive has,
even in the "privileged" sentence-initial position.) The much
smaller magnitude of the interference effect caused by the plural
possessor could be due to the fact that the possessor noun in the
first two experiments appeared before the head noun. It is possible
that interference is greater when a distracting plural intervenes
between the head noun and the verb. A third experiment was
conducted to test this possibility.
EXPERIMENT 3. The materials from Exp. 2 were modified so that
each member of the quadruplet contained a singular head noun, a
singular or plural local noun and a plural possessor noun. The
possessor modified either the head or the nonhead. Results appear
in the following table.
PREAMBLE EXAMPLES % AGREEMENT ERROR
The statue in the ELVES' garden 7%
The statue in the ELVES' GARDENS 23%
The ELVES' statue in the garden 8%
The ELVES' statue in the GARDENS 27%
There was a significant effect of local noun number. Possessor
position was not significant. These results demonstrate that plural
possessor nouns create less interference than non-specifier NPs,
independent of position in the string. It is noteworthy that even
when the possessor is the first noun in the sentence-a position
typically reserved for the subject-it has virtually no effect. In
our discussion, we consider and elaborate on possible sources for
the lesser magnitude of the effect: (1) a possessor NP, like a
pre-nominal adjective, may be less "accessible" than an NP that is
part of a modifier phrase, which could suggest that conceptual
information plays in important role in the agreement process
(Eberhard, 1999); (2) possessor NPs receive less stress than
modifier-phrase NPs (we are currently testing this possibility);
(3) agreement is computed on the basis of argument structure, not
phrase structure.
|
